>>>>> On Fri, 17 Mar 2000 11:07:32 -0500, "Will Fitzgerald" <xxxxxx@neodesic.com> said:

> Does anyone see any reason to specify, as part of the SRFI, when
> 'time' starts? Marc Feely didn't like 1/1/1900 for a starting point
> (prefers 1/1/1970), for example. The starting point clearly effects
> the implementation, or vice versa, and I'd rather give this one to
> the implementers.

I agree it should not be specified, apart from to say that the range
of times should at least include 1900.01.01 -> 2001.01.01.  Even
better I would like it to say: 1900.01.01 -> 100 years in the future.

I see no reason to base it on the 1970 Unix epoch, as you won't be
representing it as a 32 bit fixnum anyway, so you'll have to convert
regardless.

1760 might make some sense as a guaranteed minimum date.  Earlier than
that, and you get into calendar/religious questions (see the ``man
cal'' page on Unix).

../Dave